More than one in three New Zealand retail investors are now using generative AI tools such as ChatGPT and Microsoft Copilot to guide their investment decisions, according to new survey data from Chartered Accountants Australia & New Zealand (CA ANZ). The finding marks a dramatic shift in how everyday investors approach the markets, with 76% of AI users reporting they are satisfied with the outcomes. Yet beneath the surface, significant risks—from hallucinated financial figures to crowded trades—threaten to undermine the very trust that underpins market integrity.

The 2024 CA ANZ investor confidence survey, which canvassed more than 1,500 retail investors across Australia and New Zealand, revealed a nuanced picture. While 79% of Kiwi respondents expressed rising confidence in New Zealand capital markets and listed companies—a six-percentage-point jump year-on-year—confidence in overseas markets dropped by five points. AI adoption was highest among 18-to-29-year-olds (64%), and Auckland led regional usage at 51%. But the most striking number may be that 88% of investors still place their trust in audited financial statements, and auditors were ranked the most trusted group for advancing investor protection. This dual reality—eager AI experimentation paired with unwavering faith in traditional assurance—defines the current landscape.

The CA ANZ Survey: By the Numbers

The raw figures, as reported by RNZ, paint a clear picture:
- One-third (33%+) of NZ retail investors use generative AI for investment decisions.
- Satisfaction among AI users: 76%.
- Youth adoption: 64% of investors aged 18–29 use AI.
- Regional hotspots: Auckland (51%), Canterbury (33%), Wellington (27%).
- Market confidence: 79% feel more confident in NZ markets; 6-point rise year-on-year.
- Audit trust: 88% trust audited financial statements; auditors are the top-ranked guardians of market integrity.

These figures come from a multi-year CA ANZ research program, but the detailed breakdowns were provided by RNZ’s coverage. Readers should note that “use” can span from occasional ChatGPT prompts to regular Copilot-driven portfolio analysis, and the sample sizes and question phrasing may affect marginal percentage-point comparisons. Nonetheless, the trend is unmistakable: AI has become a mainstream research tool for a sizable minority of retail investors.

Why Investors Are Embracing AI

The appeal of generative AI for investing is practical and immediate. For the first time, retail investors can access research capabilities once reserved for institutional players.

  • Speed: Tools like Copilot can summarize earnings transcripts, aggregate news, and generate comparative financial summaries in minutes—collapsing hours of manual research.
  • Accessibility: Plain-language explanations of complex filings lower the barrier to entry, allowing self-directed investors to build hypothesis-driven cases.
  • Cost: Free or low-cost AI alternatives undercut expensive analyst subscriptions, particularly appealing to casual traders.
  • Democratization: AI helps screen stocks, compare metrics, and identify patterns without requiring advanced financial training.

These benefits align with broader global surveys. PwC’s Global Investor Survey, for instance, shows institutional investors expect generative AI to boost productivity significantly, and they want companies to pair AI investment with workforce upskilling. The convergence of institutional and retail sentiment underscores that AI is not a passing fad—it is reshaping investment workflows at every level.

Hidden Risks: When the Algorithm Gets It Wrong

Despite the rosy satisfaction numbers, the CA ANZ data and independent analyses expose multiple vulnerabilities that could trip up unwary investors.

1. Garbage In, Garbage Out

AI models are only as reliable as their training data. If an AI ingests stale, biased, or outright false financial data, its outputs will amplify those errors. CA ANZ and market commentators stress that audited financial statements must remain the bedrock of any analysis. Without that anchor, investors risk acting on fabricated or misleading information.

2. Hallucinations and Unverifiable Outputs

Generative models can invent plausible-sounding but entirely false claims—partnerships that don’t exist, misstated revenues, or fabricated quotes. For an investor, acting on such hallucinations could mean costly trades based on fiction.

3. Overconfidence and Fragile Trust

The survey reveals a trust gap: many non-users explicitly distrust AI, and nearly half prefer other information sources. This bifurcation suggests that heavy AI reliance without human verification could create a fragile regime where investors accept false positives until a painful correction forces a reckoning.

4. Missing Audit Trails

Most consumer AI tools lack transparent provenance and immutably logged decision trails. If a trade goes wrong, it becomes nearly impossible to reconstruct the chain of reasoning or hold any system accountable. Institutional-grade adoption demands model factsheets and reproducible test cases—features still in their infancy.

5. Crowding Risk

When thousands of investors feed similar prompts into a popular AI, the resulting trade ideas can become self-reinforcing. This concentration risk is acute in small-cap or illiquid securities, where AI-driven consensus can inflate bubbles that pop violently.

6. Regulatory Gray Zones

AI tools that drift from “information” to “advice” may inadvertently cross into regulated financial advice territory. Jurisdictional rules are nascent, and the distinction will be litigated in coming years, exposing users and developers to compliance ambiguity.

The Auditor as Data Guardian

Perhaps the most forward-looking theme to emerge from the CA ANZ findings is the reframing of auditors and accountants as “data guardians.” With 88% trust in audited financials and auditors crowned the most trusted protectors of market integrity, the profession has a unique opportunity to step into a new role: certifying the inputs that feed AI-driven investment systems.

CA ANZ explicitly ties the continued relevance of audit to AI adoption. A parallel study from Chartered Accountants Worldwide/Ipsos indicates that younger accountants are heavy AI users themselves, and the wider profession sees mounting responsibility for governing the information pipelines that sustain investor-facing AI. The logical next step is the development of assurance protocols, model factsheets, and data lineage certifications that give investors confidence in the “guts” of their AI tools.

Practical Playbook for Retail Investors

For investors already using or considering AI, a few disciplined habits can dramatically reduce downside risk:

  • Verify everything: Cross-check any AI-generated numerical claim (revenue, margins, contract terms) against primary filings or audited statements before acting.
  • Use AI as a research scout, not an execution engine: Maintain human oversight for every buy/sell decision.
  • Demand provenance: Favor tools that show source links, timestamps, and confidence scores.
  • Keep a decision log: Save prompts, outputs, and verification steps as a contemporaneous record.
  • Understand data freshness: Know whether the AI relies on cached data, real-time feeds, or proprietary datasets.
  • Size conservatively: When acting on AI-sourced signals, reduce position sizes to account for elevated uncertainty.

What Firms, Advisors, and Regulators Must Do

The AI genie is out of the bottle, but that doesn’t mean anarchy. Coordinated action across the investment ecosystem can channel AI’s potential while preserving trust.

For platforms and advisors:
- Integrate authenticated data sources (exchange feeds, audited filings) and publish model factsheets.
- Build immutable audit trails for prompts, outputs, and data provenance.
- Embed human-in-the-loop approvals for high-impact decisions.

For regulators and standard-setters:
- Draft guidance that clearly separates information tools from regulated financial advice.
- Require minimum disclosure standards for AI-driven recommendation services.
- Fund investor AI literacy programs to reduce asymmetrical comprehension.

For auditors:
- Expand assurance frameworks to cover AI-fed analytics and the datasets used to train investment models.
- Certify the data pipelines and model governance processes that underpin AI tools.

Synthesis: Opportunity with Clear Boundaries

The CA ANZ survey signals an inflection point. Retail investors are not just experimenting with AI—they are embedding it into their decision-making, and most report positive experiences. This democratization of analytical firepower is a market development to be celebrated, but it does not eliminate the need for reliable, auditable source data.

Auditors and accountants are poised to become the linchpin of trust, stepping into the data guardian role by certifying the inputs that feed investor-facing AI. Platforms must concurrently improve provenance, transparency, and explainability to convert the fragile trust of early adopters into durable confidence.

The practical path forward is clear: embrace AI as an augmentative tool, not a final decision-maker; insist on verifiable sources and auditable trails; and always use human judgment as the final arbiter for actions that materially affect wealth. When these guardrails are in place, AI can be a powerful productivity enhancer for investors. Without them, the same technology risks amplifying errors and producing concentrated, fragile market behaviors that hurt the very investors it aims to empower.